Erik McClure

extremely amused by recruiters just straight-up admitting their client has given them impossible requirements that NOBODY qualifies for so they'll take anyone who even gets vaguely close to them (which is why they are talking to me).

We need a movie about building software where every single inspiring montage about development ends with a smash cut to "wait, this doesn't work at all. Fuck." followed by weeks of desperately trying to find a new architecture that isn't fundamentally broken, and at the end of the movie the entire scene where they do the final release plays out in complete silence, with no music other than the tapping of keyboards, followed by "...so what do we do now?" [CREDITS]

this would not be a fun movie to watch I just want people to know my pain

Tried to use C++23 for all of two days and instantly found a compiler bug in MSVC and libstdcxx

how does this keep happening

@interru@hooves.social My understanding is that ActivityPub doesn't standardize the exported data, therefore making a general solution impossible unless everyone at least agrees on a de-facto standard. If that happened, then sure, you could make it work, so uh, I guess lemme know when that happens.

Apparently there are rumors about Bluesky enshittifying because the CEO resigned or something, and I don't think this will happen, but the thing that's tiring about migrating between platforms is that you lose all your data. NONE OF THE PROPOSED ALTERNATIVES ACTUALLY FIX THIS PROBLEM. Mastodon lets you redirect your old handle to your new one and then tells you to go fuck yourself. ActivityPub, in general, is architecturally incapable of doing this. The only protocol that could even theoretically do this is... ATProto. From Bluesky. But only in theory, because in practice the main bsky server controls your DID so you still need their permission to actually migrate the DID to a different PDS, and AFAIK this is still not automated or exposed by their API. But also if I point out that bluesky has the only protocol that could *even theoretically* do this, a bunch of terminally online activists will call me a capitalist for no reason and demand I apologize for wanting nice things.

RE: @mttaggart@infosec.exchange

We desperately need a three strikes system of some kind. Like, you get *at least* two free bad takes before you're considered problematic. Nobody is perfect.

Furthermore, this has a basis in game theory, as Nicky Case demonstrated with the "forgiving" tit-for-tat strategy that was successful in the iterated prisoner's dilemma when there was a 1% chance of making a mistake: https://ncase.me/trust/

I'm so tired of being unable to critique or defend something without all of my nuance being pissed on. I'm going to respond violently to the next person taking my logical statements as tribal signaling.

I had a bug, so I broke up complex operations into small ones and made the types more strict so I could verify the correct order of operations at compile time. Once I got it compiling again, the bug was fixed. If we just used types properly we could convert 95% of runtime bugs into type errors. We have all the tools we need, but the software industry continues to ship buggy garbage because we refuse to spend time making our code harder to misuse and instead spend all that time in a debugger trying to figure out where we forgot to add something. You don't even need fancy type systems for this, you can do it in Rust or C++. Anything that lets you wrap the result of a computation in a transparent type that marks the stage of the computation.

the parmesan cheese was clumping up so I shook the container to break it up and apparently it wasn't closed all the way so I am now slightly dusted with parmesan cheese

I'm seeing posts that say "a country that can fund wars but not healthcare has a values problem", and sure, this is true, but people were saying the EXACT SAME THING 20 years ago when I was in high school. Clearly, just SAYING things isn't enough. We've been saying it for decades. Nothing changed. Just telling people they have terrible morals clearly doesn't do anything. As long as there are incentives to starting wars and cutting benefits, we will start wars and cut benefits. As long as there are incentives to being a horrible person, people will find a way to morally justify even the most heinous acts. Appealing to people's sense of morality almost never works. I'm tired of hearing it. If you want people to stop doing bad things, change how the system works so they're not incentivized to do bad things. If you don't know how to do that, maybe figure that part out BEFORE you go around complaining about people behaving like rational agents in a low-trust environment.

RE: https://chaos.social/@gsuberland/116183332718393884 more people need to realize that bad faith actors will use any means necessary to launder their shitty behavior into society.

it was easier to enjoy things when I didn't know everything was giving me cancer

RE: https://meow.social/@novra/116136959005134188 I'll have the last laugh when I outlive them as a 150 year old cyborg.

Thinking about how, despite the fact that I've almost never had a "bad brain day" and have no clue what an anxiety or panic attack is like, when people tell me they have these problems, I just... listen to them? My ADHD friend didn't get work done today because it was a bad brain day? Okay, cool, we'll try again tomorrow. Meanwhile the rest of society is like "WOW THESE PEOPLE MUST BE SO PATHETIC AND LAZY AND WORTHLESS" w h a t ? ? ? ? Nothing feels more alienating than to watch *multiple human societies* made of apex omnivores that supposedly share my genetics decide to torture random collections of other humans for obviously insane reasons, over and over and over. Sometimes it feels like humans have an easier time understanding absurd scifi aliens than each other. If you're sufficiently alien, at some completely arbitrary point it suddenly becomes okay to be different. Maybe that's why I wish I really was an alien.

once again, i have written an architecture that can only be properly debugged using a graph. if you're wondering, there shouldn't be any green nodes underneath a yellow node. Yes, that means something is very broken.

RE: https://yiff.life/@erkhyan/116127756821817705 this took way too long, but at least it happened

"that is an impressive collection of incorrect statements" is my new favorite insult

RE: https://wandering.shop/@silvermoon82/116120882785675157 I think this situation is dangerously plausible, precisely because I think LLMs will be legitimately useful for certain small tasks in the future, like refactoring code or generating certain kinds of boilerplate. This is actually WORSE than there being *no* legitimate use cases of LLMs, because it will provide a permanent veneer of viability that will be used to justify unlimited subsidies on the premise that *next year* they'll finally be able to make the LLM write an entire app by itself, trust us it'll happen this time just give us one more year of subsidies bro please bro and this will just go on forever until we supplant LLMs with a different paradigm or there's a butlerian jihad or the planet explodes or whatever.

It's either very funny or very depressing to watch executives trip over themselves to prove who has the worst understanding of what software development actually entails.

https://www.tumblr.com/derinthescarletpescatarian/808572526094598145/on-fun-thing-about-everyone-using-ai-to-do-their

minecraft liminal industries https://www.twitch.tv/cloudhop

I know many things about floats, like how you should always take the absolute value before doing a bitwise comparison to 0 so a negative zero doesn't sneak in, or that different NaN values can have different mantissa payloads under bitwise comparisons. What I didn't know, is that the floating point additive identity, when taking into account bitwise comparisons, is NEGATIVE ZERO, not positive zero, which means zeroing a struct with floats in it technically doesn't give you the additive identity??? This is because (-0.0) + (+0.0) == (+0.0) but (-0.0) + (-0.0) == (-0.0)!!! Obviously this only matters under bitwise comparison since +0.0 equal to -0.0 under float comparisons, but wow I was really surprised by this one. https://users.rust-lang.org/t/why-is-the-additive-identity-of-floats-0-0/128016

This article is fantastic because it's been a long time since I've seen someone do proper statistical analysis. It seems questionable at the start, but just keep reading, it makes sense in the end, and it explains how to avoid a bunch of data analysis fallacies in the process. https://campedersen.com/singularity

factoriowoing (in minecraft) https://www.twitch.tv/cloudhop

This post is definitely right about "notbyai.fyi" being corporate gaslighting, but I'm increasingly alarmed at leftists treating "the advancement of humanity" as a red flag simply because it is echoed by a bunch of techbros. Letting fascists poison an entire concept is a memetic trap. You are conceding the position in advance, even if that's not what you intended. Solarpunk is also "pushing humanity forward", just in a better direction. Don't let the empty promises of a fascist undermine people's genuine desire to improve society. Pushing humanity forward means destroying the status quo. Failing to destroy the status quo means death. https://social.treehouse.systems/@imbl/115978426251286619

LETS JUMP OFF A CLIFF! https://www.twitch.tv/cloudhop

minceraft backdooms: https://www.twitch.tv/cloudhop

@jk@mastodon.social it'll definitely run, but it won't actually work properly, which is the perpetual issue with opengl

@ricardoharvin@mstdn.social I was already mentally prepared to experience the slow collapse of the entire planet's biosphere over the next 50 years though. It's priced in.

When I promised myself that I would continue having new experiences even as I got older, I was thinking more along the lines of "learn new skills and see new places", not "experience nazi germany and the dot-com bubble at the same time"

damn ya'll I wrote some fucking bangers when I was younger, I just didn't have the sound design needed to make them work

‪This is very similar to a blog post I wanted to write about, on how an extremely intelligent person can construct a plausible argument in favor of almost anything, which is a problem, because if no one else is able to figure out why their arguments don't work, it becomes impossible to tell what is or isn't plausible, and it starts to seem like everything is plausible, and then they go insane. https://www.joanwestenberg.com/why-intelligence-is-a-terrible-proxy-for-wisdom/

Bose recently did an unambiguously good thing, by publishing the API for the audio hardware they were originally going to brick: https://www.theverge.com/news/858501/bose-soundtouch-smart-speakers-open-source

However, I've seen some people say "don't praise Bose for this, they didn't do this until there was backlash".

SHUT UP. Shut the FUCK UP. I'm DONE living in a society where you get dragged through hell if you make a mistake, EVEN AFTER YOU CORRECT THE MISTAKE. I'm so fucking tired of hearing stupid excuses for this kind of puritanism like "they should've known better" NOBODY KNOWS BETTER UNTIL *AFTER THEY MAKE THE MISTAKE*. THAT'S HOW LEARNING *WORKS*.

And before you say "Companies aren't your friend" PUNISHING THEM FOR FIXING THEIR MISTAKES WON'T MAKE THEM DO THE RIGHT THING EITHER. If other people, or companies, see someone get punished for both messing up AND fixing the mistake, they just won't bother at all!

People HAVE to be allowed to make mistakes. They HAVE to be given a chance to improve.